This server enables AI assistants to manage Featureflow feature flags, projects, and environments programmatically through the Model Context Protocol.
Core Capabilities:
Account Management: Get current user and organization information
Project Management: List, create, view, update, and delete projects with search filtering
Feature Flag Management: Full lifecycle management including create, update, clone, archive/unarchive, and delete with filters for project, archived status, and predefined filters (maintaining, bookmarked, recent)
Feature Control Configuration: View and modify feature settings per environment including enable/disable states, off variants, and targeting rules
Environment Management: Create, list, update, and delete environments with cloning capabilities and safeguards
Targeting System: List and view targeting attributes (user attributes) used for A/B testing rules
API Key Management: List SDK API keys for environments (server and client types)
Self-Hosted Support: Configure for self-hosted Featureflow instances
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Featureflow MCP Server enable the 'dark-mode' feature in production"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
An MCP (Model Context Protocol) server for Featureflow feature flag management. This enables AI assistants like Claude to interact with your Featureflow instance to manage feature flags, projects, environments, and more.
Log into Featureflow as an administrator
Navigate to Administration → API Tokens
Click Create Token and copy the token (starts with api-)
Add to your ~/.cursor/mcp.json:
{
"mcpServers": {
"featureflow": {
"command": "npx",
"args": ["-y", "featureflow-mcp"],
"env": {
"FEATUREFLOW_API_TOKEN": "api-your-token-here"
}
}
}
}Press Cmd+Shift+P → "MCP: Restart Servers" or restart Cursor.
That's it! You can now ask Claude to manage your feature flags.
Related MCP server: Unleash MCP Server
Environment Variable | Description | Default |
| Personal Access Token (required) | - |
| API base URL (optional) |
|
If you're running a self-hosted Featureflow instance:
{
"mcpServers": {
"featureflow": {
"command": "npx",
"args": ["-y", "featureflow-mcp"],
"env": {
"FEATUREFLOW_API_URL": "https://your-instance.com/api",
"FEATUREFLOW_API_TOKEN": "api-your-token-here"
}
}
}
}Tool | Description |
| Get the currently authenticated user and organization |
Tool | Description |
| List all projects, optionally filtered by query |
| Get a specific project by ID or key |
| Create a new project |
| Update an existing project |
| Delete a project |
Tool | Description |
| List features with optional filters |
| Get a specific feature by ID or unified key |
| Create a new feature flag |
| Update an existing feature |
| Clone a feature with a new key |
| Archive or unarchive a feature |
| Delete a feature |
Tool | Description |
| Get feature control settings for an environment |
| Enable/disable features, modify rules |
Tool | Description |
| List environments for a project |
| Get a specific environment |
| Create a new environment |
| Update an existing environment |
| Delete an environment |
Tool | Description |
| List targeting attributes for a project |
| Get a specific target by key |
| List SDK API keys for an environment |
Once configured, you can ask Claude things like:
"Who am I logged in as in Featureflow?"
"List all my Featureflow projects"
"Create a feature called 'new-checkout' in the 'webapp' project"
"Enable the 'dark-mode' feature in production"
"What features are currently enabled in staging?"
"Disable 'beta-feature' in all environments"
# Clone the repository
git clone https://github.com/featureflow/featureflow-mcp.git
cd featureflow-mcp
# Install dependencies
npm install
# Build
npm run build
# Run locally
FEATUREFLOW_API_TOKEN=api-xxx npm startMIT - see LICENSE for details.
Featureflow - Feature flag management platform
MCP Protocol - Model Context Protocol specification
Featureflow Documentation - API documentation
Archive or unarchive a feature flag.
Inputs: idOrUnifiedKey (required), archived (required).
Clone an existing feature with a new key and name.
Inputs: idOrUnifiedKey (required), newKey (required), name (required).
Create a new environment for a project. Optionally clone settings from an existing environment.
Inputs: projectKey (required), key (required), name (required), color, +2 more.
Create a new feature flag in a project.
Inputs: projectKey (required), key (required), name (required), description.
Create a new project in the organization.
Inputs: key (required), name (required).
Delete an environment. Cannot delete the last environment in a project.
Inputs: idOrUnifiedKey (required).
Delete a feature flag. Requires production editor or admin permissions.
Inputs: idOrUnifiedKey (required).
Delete a project. This will also delete all features and environments in the project.
Inputs: idOrKey (required).
Get detailed information about a specific environment by ID or unified key (projectKey:environmentKey).
Inputs: idOrUnifiedKey (required).
Get detailed information about a specific feature by ID or unified key (projectKey:featureKey).
Inputs: idOrUnifiedKey (required).
Get the feature control configuration for a specific feature and environment. Shows enabled state, rules, and variant assignments.
Inputs: idOrUnifiedKey (required), environmentKey (required).
Get detailed information about a specific project by its ID or key.
Inputs: idOrKey (required).
Get a specific target by its key.
Inputs: projectKey (required), targetKey (required).
List API keys for a specific environment.
Inputs: environmentKey (required), type.
List all environments for a project or the entire organization.
Inputs: projectKey.
List all features. Can filter by project key, search query, or predefined filters (maintaining, bookmarked, recent).
Inputs: projectKey, query, filter, archived.
List all projects in the organization. Optionally filter by a search query that matches project name or key.
Inputs: query.
Get all targets (user attributes) for a project. Targets are used in targeting rules for A/B testing.
Inputs: projectKey (required).
Update an existing environment's properties.
Inputs: idOrUnifiedKey (required), name, color, url, +1 more.
Update an existing feature's properties like name, description, or variants.
Inputs: idOrUnifiedKey (required), name, description.
Update feature control settings for a specific environment. Can enable/disable the feature, change the off variant, and modify rules.
Inputs: idOrUnifiedKey (required), environmentKey (required), enabled, offVariantKey.
Update an existing project's name.
Inputs: idOrKey (required), name (required).
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 22 tools
Each tool has a clearly distinct purpose with no ambiguity, as they target specific resources (feature, environment, project, target, API key) and actions (create, get, list, update, delete, clone, archive). Overlap is minimal, such as 'get_feature' and 'get_feature_control' which serve different levels of detail, and descriptions clarify boundaries effectively.
Tool names follow a highly consistent verb_noun pattern throughout, using snake_case uniformly (e.g., create_feature, list_environments, update_project). The naming convention is predictable, making it easy for agents to infer functionality and maintain readability across all 22 tools.
With 22 tools, the count is slightly high but reasonable for a feature flag management system, covering CRUD operations for multiple resources (features, environments, projects, targets, API keys). It feels comprehensive rather than bloated, though it borders on the upper limit of typical scoping (3-15 tools).
The tool surface provides complete CRUD/lifecycle coverage for the feature flag management domain, including creation, retrieval, listing, updating, deletion, and specialized operations like cloning and archiving. No obvious gaps exist; agents can perform all core workflows without dead ends, from project setup to feature control updates.
Manage ConfigDirector feature flags, remote configs, and targeting rules from AI assistants.
Evaluate and inspect your Hypertune feature flags from your AI assistant.
Plan Salesforce deploys, open pull requests and trigger pipelines from your AI client.
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
